Evaluating Temporal Plasticity in Foundation Time Series Models for Incremental Fine-tuning
This addresses the problem of enabling continuous learning in time series models for applications with evolving data distributions, though it appears incremental in methodology.
The paper investigated whether time series foundation models can improve continuously through incremental fine-tuning while maintaining existing capabilities, finding that foundation models like Time-MoE and Chronos showed sustained accuracy improvements while traditional models deteriorated.
Time series foundation models excel at diverse time series forecasting tasks, but their capacity for continuous improvement through incremental learning remains unexplored. We present the first comprehensive study investigating these models' temporal plasticity - their ability to progressively enhance performance through continual learning while maintaining existing capabilities. Through experiments on real-world datasets exhibiting distribution shifts, we evaluate both conventional deep learning models and foundation models using a novel continual learning framework. Our findings reveal that while traditional models struggle with performance deterioration during incremental fine-tuning, foundation models like Time-MoE and Chronos demonstrate sustained improvement in predictive accuracy. This suggests that optimizing foundation model fine-tuning strategies may be more valuable than developing domain-specific small models. Our research introduces new evaluation methodologies and insights for developing foundation time series models with robust continuous learning capabilities.